中文摘要：

转炉终渣成分对后续炼钢过程将产生重要影响.采用Visual Basic 6.0进行编程,应用克服BP神经网络缺陷的小波神经网络,建立了网络结构为8-10-6,其中隐含层传递函数为Morlet型函数,输出层传递函数为S型函数的120 t转炉终渣成分预报模型.采用550炉数据进行模型训练,经20炉数据现场验证表明,模型预报结果各个成分有85.8％的平均相对误差在0.1以内.模型预测精度较高,可以满足工厂实际使用需要.

英文摘要：

The composition of the terminal slag has important effect on the following operation of steelmaking. By using Visual Basic 6 software, application of wavelet neural network to overcome the shortcomings of BP neural network, the prediction model of the terminal slag composition in an 120 t Converter were developed. The network structure of the model was 8-10-6, the implied layer transfer function of the model was a Morlet function and the output layer transfer function was S function. The training data and the prediction data for the model were 550 and 20 heats respectively. The results showed that the average relative error of 85.8 percent heats was within 0.1. The model has high forecast precision; it could meet the requirement of the steel company.